Patient-Specific iPSC-Derived Astrocytes Contribute to Non-Cell-Autonomous Neurodegeneration in Parkinson's Disease

1Parkinson's disease (PD) is the second most prevalent neurodegenerative disease after Alzheimer's disease, affecting 7 to 10 million people worldwide. PD is characterized by a significant loss of ventral midbrain dopaminergic neurons (vmDAns) in the substantia nigra pars compacta. The presence of intracellular protein aggregates of α-synuclein (α-syn) in the surviving vmDAns has been reported in postmortem PD tissue. Most PD cases are sporadic (85%), but familial mutations are accountable for 15% of patients. Mutations in the gene encoding leucine-rich repeat kinase 2 (LRRK2), causing an autosomal dominant form of PD, account for 5% of familial cases and 2% of sporadic cases. LRRK2 is a highly complex protein with both GTPase and protein kinase domains involved in several cellular functions, including autophagy.

2Correlations between mutant LRRK2 and several pathogenic mechanisms linked to PD progression have been previously reported, including alterations in autophagy and consequent accumulation of α-syn. Neuronal mutant LRRK2 toxicity was found to depend on LRRK2 levels and α-syn accumulation as opposed to kinase activity or inclusion bodies in induced pluripotent stem cell (iPSC)-derived neurons. During PD pathogenesis, mutant LRRK2 was found to directly bind LAMP2A, the receptor responsible for chaperone-mediated autophagy (CMA) normally used by both LRRK2 and α-syn for degradation. This binding blocks the proper functioning of the CMA translocation complex, resulting in defective CMA, leading to the accumulation of α-syn and cell death.

3iPSCs derived from healthy individuals and patients have accelerated advances in developing genuinely human experimental models of diseases. In the case of PD, previous studies by our groups and others have generated iPSCs from patients with PD associated with LRRK2 mutations, and described the appearance of disease-specific phenotypes in iPSC-derived neurons, including impaired axonal outgrowth and deficient autophagic vacuole clearance. Moreover, dopaminergic (DA) neurons from LRRK2-mutant iPSCs displayed alterations in CMA that were, at least in part, responsible for the abnormal accumulation of α-syn observed in these cells, which predated any morphological signs of neurodegeneration.

4Studies investigating PD pathogenesis have been mostly focused on the mechanisms underlying vmDAn degeneration and death. However, there is evidence of astrocytes accumulating α-syn during PD through postmortem analysis. Altered α-syn released by axon terminals in the surrounding synapses is taken up by astrocytes, supporting the hypothesis of the spread of α-syn through neuron-astrocyte interactions. Overexpression of mutant SNCA in primary astrocytes altered their normal functioning and impaired proper blood-brain barrier control and glutamate homeostasis, and eventually resulted in a significant loss of vmDAns. In a study using human brain homogenates from PD patients with Lewy bodies, α-syn was found to be taken up and spread from astrocytes to neurons, leading to neuronal death. As a result, a role of astrocyte dysfunction in PD pathogenesis is emerging.

5In the present study, we generated patient-specific iPSC-derived astrocytes and vmDAns from PD patients with the LRRK2 G2019S mutation and healthy individuals. We consistently generated a population of human vmDAns in vitro that expressed postmitotic dopaminergic markers and fired action potentials. Subsequently, we co-cultured healthy iPSC-derived vmDAns with iPSC-derived astrocytes expressing the mutated form of LRRK2 associated with PD. In co-culture experiments, we detected a significant decrease in the number of vmDAns in the presence of LRRK2-PD astrocytes, which correlated with the abnormal accumulation of astrocyte-derived α-syn. Conversely, control astrocytes were able to partially rescue disease-related phenotypes in LRRK2-PD vmDAns during co-culture. A more in-depth investigation revealed impaired autophagic machinery, as well as progressive accumulation of endogenous α-syn in PD astrocytes, compared with control astrocytes. By treating the cells with an activator of CMA, we were able to prevent the appearance of PD-related phenotypes in patients' astrocytes. Overall, our findings represent a direct indication that dysfunctional astrocytes play a crucial role during PD pathogenesis and may have broad implications for future intervention in early stages of PD.

6To establish an in vitro human cellular model for dissecting the interplay between neurons and astrocytes in PD, we first derived astrocyte-like cells from iPSCs, using a previously published protocol. Specifically, astrocyte cultures were successfully established from iPSC lines from three PD patients carrying the G2019S mutation in the LRRK2 gene (PD SP06, PD SP12, and PD SP13) and two healthy age-matched controls (Ctrl SP09 and Ctrl SP17) (see Table 1 and Tables S1 and S2 for a summary of the iPSC lines used, and Experimental Procedures for details on their origin). Immunocytochemistry (ICC) detection of key astrocyte markers showed robust expression of CD44, glial fibrillary acidic protein (GFAP), and S100 calcium-binding protein β (S100β), as well as of the excitatory amino acid transporter 2 (EAAT2, also known as GLT1), in all human iPSC-derived astrocytes (Figure 1A). No evident contamination by other cell types, such as neurons or oligodendroglial progenitors, was found as assessed by immunostaining with anti-MAP2 or NG2 antibody, respectively (Figures 1A and 1B). The astrocytic identity was further confirmed by quantitative RT-PCR of GFAP and additional astrocyte-specific genes, including MLC1, SOX9, ALDH1L1, AQP4, DIO2, and SLC4A4, which were expressed in Ctrl and PD astrocytes, and in human primary astrocytes, but not in iPSCs (Figures S1A–S1C).

7To validate astrocyte cell type identity, we also tested the expression of astrocyte-specific genes revealed by the Human Astrocyte RNA-Seq database (www.brainrnaseq.org/) in our iPSC-derived astrocytes, through RNA sequencing. We found that the transcriptomic profile of both Ctrl and PD iPSC-derived astrocytes was closer to that of human primary astrocytes than to that of their corresponding iPSC line, thus confirming their identity (Figure 1C). We next determined the functional maturation of iPSC-derived astrocytes by confirming their capacity to produce ATP (Figure 1D) and propagate intercellular Ca2+ waves (Figures S1D–S1H). Indeed, by using the Fluo-4 AM Ca2+ indicator, recordings from Ctrl and PD astrocytes showed a heterogeneous pattern of Ca2+ fluctuations under basal conditions, revealing their functionality. All together, these data support the successful generation of highly pure populations of functionally equivalent astrocyte-like cells, which represent a continuous source of human Ctrl and PD astrocytes for subsequent analyses.

8To investigate whether astrocytes contribute to PD pathogenesis, we established a co-culture system of iPSC-derived astrocytes and iPSC-derived vmDAns. PD vmDAns were generated from iPSC lines from two PD patients carrying the G2019S mutation in the LRRK2 gene (iPSC lines PD SP12 and PD SP13), whereas Ctrl vmDAns were obtained from two independent iPSC clones (Ctrl SP11 and Ctrl SP11#4) from a healthy age-matched control (see Tables 1, S1, and S2 for a summary of the iPSC lines used). To differentiate iPSCs toward vmDAns, we used a combination of two previously published midbrain floor-plate differentiation protocols that was comparably effective in all iPSC lines analyzed. Under these conditions, ∼20% of cells in the cultures were committed to DA neuron fate by day 35 of differentiation, as judged by the expression of tyrosine hydroxylase (TH) and forkhead box protein A2 (FOXA2) (Figure 2A). By day 50 of differentiation, the percentage of TH+ neurons reached ∼30%, most of which also expressed the A9-domain-specific marker G-protein-activated inward rectifier potassium channel 2 (GIRK2), and displayed spontaneous action potential firing (Figures S2B–S2D). For co-culture experiments with control astrocytes, we dissociated vmDAn cultures after 35 days of differentiation and plated them onto a confluent layer of Ctrl iPSC-derived astrocytes (Figure 2E). After 4 weeks of co-culture, we found that astrocyte-neuron glutamate exchange was present through glutamate transporter 1 (GLT1) expression (Figure 2F) and neuronal synapse formation (Figure 2G). Accordingly, an overall healthy neuronal network comprising MAP2-positive cells was formed upon co-culture (Figure 2H).

9We then examined the effects of astrocytes expressing mutated LRRK2 on the survival of Ctrl iPSC-derived vmDAns upon co-culture (Figure 3A). After 2 weeks of co-culture with PD astrocytes, Ctrl vmDAns displayed morphological alterations, including shortened neurites, and significantly decreased cell survival compared with co-cultures with Ctrl astrocytes (Figures S2A–S2C). These alterations were much more evident after 4 weeks of co-culture, when Ctrl vmDAns cultured on top of PD astrocytes showed extensive signs of neurodegenerative phenotypes (fewer and shorter neurites, and abundance of beaded-necklace neurites) and severely compromised cell survival (less than ∼25% of control), compared with co-cultures with control astrocytes (Figures 3B–3F). The fact that the numbers of vmDAns did not change significantly during the co-culture with control astrocytes, but progressively declined when co-cultured with PD astrocytes, strongly suggests that vmDAns were lost under the latter conditions as a result of neurodegeneration, rather than a blockade in vmDAn differentiation or maturation.

10Viability tests of both Ctrl and PD astrocytes at 2 and 4 weeks of co-culture revealed highly similar values, indicating that neurodegenerative signs displayed by Ctrl vmDAns were not caused by a dying PD astrocyte (Figure S2D). Interestingly, vmDAn neurodegeneration upon co-culture with PD astrocytes was specific to this type of neuron, because non-dopaminergic neurons (TH−/MAP2+) did not significantly change in numbers or morphology after co-culture with Ctrl or PD astrocytes (Figures S2E–S1J). All together, these results indicate a neurotoxic capacity of PD astrocytes toward vmDAns, with no effects on other neuronal types concomitantly present in cultures.

11Given the relevance of α-syn in the context of PD pathogenesis, we sought to examine whether vmDAns co-cultured with PD astrocytes abnormally accumulated α-syn. α-syn was barely detectable in the cytoplasm of Ctrl vmDAns cultured alone (data not shown) or when co-cultured with Ctrl astrocytes (Figure 3G). In contrast, Ctrl vmDAns accumulated high levels of α-syn throughout the neurites and cell body after 4 weeks of co-culture with PD astrocytes (Figures 3H and 3I). Notably, while Ctrl astrocytes had undetectable levels of α-syn (Figure 3J), PD astrocytes displayed high levels of α-syn (Figures 3K and 3L), raising the intriguing possibility that α-syn from PD astrocytes might be transferred to Ctrl vmDAns. To directly address if this was the case, we genetically engineered two iPSC lines (representing one PD patient and one healthy control) using CRISPR/Cas9 technology so that the endogenous α-syn would be tagged with a FLAG peptide (α-syn-FLAG iPSC lines; Figures S2K and S2L and Tables S2 and S3). α-syn-FLAG-tagged astrocytes were generated and fully characterized (Figure S2M). As expected, PD α-syn-FLAG tagged astrocytes accumulated abnormally high levels of α-syn, which co-localized with anti-FLAG staining (Figure 3M). More importantly, the co-culture of Ctrl vmDAns on top of α-syn-FLAG-tagged PD astrocytes for 4 weeks resulted in FLAG-tagged α-syn accumulation in neurons, demonstrating the direct transfer of astrocytic α-syn to neurons (Figure 3N).

12In addition to co-culturing cells with direct glia-neuron contact, we tested the effect of culturing Ctrl vmDAns with medium conditioned by Ctrl or PD astrocytes at different concentrations (Figure S3A). Exposure of Ctrl vmDAns to PD astrocyte-conditioned medium for 1 week, even at low concentrations, resulted in α-syn accumulation, morphological alterations suggestive of neurodegeneration, and decreased cell survival (Figures S3B–S3H), indicating that PD astrocytes secrete a molecule(s) that is toxic to vmDAns. Direct uptake by vmDAns of α-syn from conditioned medium was tested by exposing Ctrl vmDAns to medium collected from PD α-syn-FLAG-tagged astrocytes (Figures S3I–S3J), suggesting that the neurotoxic effect of PD astrocytes on vmDAns is, at least in part, mediated by secretion of α-syn.

13We have previously shown that vmDAns derived from PD-iPSCs show signs of neurodegeneration (including reduced numbers of neurites and neurite arborization, as well as accumulation of abnormal α-syn in the soma) after 50 days of culture, which are not evident in Ctrl vmDAns. To test whether the neurodegeneration could be rescued or prevented by healthy astrocytes, we co-cultured PD vmDAns with Ctrl or PD astrocytes. After a 4-week co-culture, PD vmDAns showed a partially recovered neurite number and complex neurite arborization when co-cultured on control astrocytes, compared with co-cultures with PD astrocytes (Figures 4A–4E). This rescue was partial, since PD vmDAns co-cultured with control astrocytes did not reach the levels of cell survival and complex neurite arborization seen in co-cultures of Ctrl vmDAns and Ctrl astrocytes (compare Figures 4A and 4C–4E with 3B and 4D–4F). Moreover, co-culture with Ctrl astrocytes also prevented the accumulation of α-syn in PD vmDAns that was evident in co-cultures with PD astrocytes (Figures 4A, 4B, and 4F). Notably, most Ctrl astrocytes when co-cultured with PD vmDAns adopted a flat morphology with moderate levels of α-syn; however, some harbored a hypertrophic morphology with retracted processes that accumulated very high levels of α-syn (Figures 4G and 4H), suggesting that reactive astrocytes may contribute to the clearance of vmDAn α-syn accumulation. Culture of PD vmDAns with medium conditioned by Ctrl astrocytes also rescued cell survival, morphological alterations, and α-syn accumulation (Figures S3K–S3S), indicating that direct neuronal-glial contact was not necessary for the neuroprotective effect.

14We next investigated the causative role of the genetic background of patient-specific astrocyte cells by ectopically expressing mutated LRRK2 G2019S in Ctrl astrocytes. In these experiments, Ctrl astrocytes were transfected with a plasmid expressing V5-tagged LRRK2 G2019S, or with a Ctrl plasmid expressing GFP, and analyzed 7 days after transfection for the presence of α-syn. Astrocytes transfected with LRRK2 G2019S exhibited diffuse cytoplasmic accumulations of α-syn (Figure S4A), which were not present in GFP-transfected cells (Figure S4B). The transfection efficiencies (30%–40%, as evaluated by co-staining for V5/GFAP or GFP/GFAP) were comparable under both conditions (Figure S4C). Next, we co-cultured Ctrl vmDAns for 4 weeks with LRRK2 G2019S-transfected Ctrl astrocytes, or with GFP-transfected astrocytes as a control, and we found α-syn accumulation in 50% of the TH+ neurons only in co-cultures with LRRK2 G2019S-transfected astrocytes (Figures S4D–S4G). Overall in these cultures we found decreased survival of vmDAns and evident morphological alterations (Figures S4H–S4I), including fewer and shorter neurites compared with vmDAns cultured on top of GFP-transfected astrocytes, indicating that the expression of pathogenic LRRK2 in Ctrl astrocytes is deleterious for the survival of dopaminergic neurons.

15For the converse experiment, we generated isogenic PD astrocytes lacking the LRRK2 G2019S mutation by CRISPR/Cas9-mediated gene editing of PD iPSCs (iPSC line PD SP13, from here on referred to as PD iso), and fully characterized these cells (Figures S5A–S5E). Abnormal α-syn accumulation did not occur in gene-corrected astrocytes, in contrast with their isogenic mutant counterparts (Figures S5F and S5G). Moreover, co-culturing gene-corrected astrocytes with Ctrl vmDAns for 4 weeks prevented the accumulation of α-syn and decrease in neuron survival observed when Ctrl vmDAns were co-cultured with PD astrocytes (Figures S5H–S5J), further supporting that the expression of mutant LRRK2 in astrocytes is pathogenic to Ctrl vmDAns.

16Since PD astrocytes displayed higher levels of α-syn compared with controls, we next investigated possible differences in α-syn turnover in these cells. Degradation of α-syn in lysosomes occurs in large extent through CMA. To investigate possible changes in CMA in PD astrocytes, we first stained at 6 and 14 days for both α-syn and LAMP2A, the receptor for CMA (Figures 5A and S6A). Ctrl astrocytes showed LAMP2A in the perinuclear area (perinuclear lysosomal positioning occurs during CMA activation; ) and barely detectable levels of α-syn at both 6 and 14 days (Figures 5A, 5B, and S6A). In contrast, PD astrocytes displayed LAMP2A-positive vesicles all around the cell body as early as 6 days, which continued to be present after 14 days (Figures 5A, 5B, and S6A). Moreover, abnormal accumulation of α-syn was confirmed in PD astrocytes after 14 days of culture, compared with Ctrl astrocytes (Figures 5A–5E). Interestingly, this accumulation was not present after 6 days of culture, suggesting progressive α-syn accumulation over the 14-day time period. Co-localization analyses of α-syn with the LAMP2A receptor revealed a positive co-localization that was more evident in PD astrocytes (Figures 5C and S6B). CMA substrates are usually rapidly internalized and degraded inside lysosomes, but we have previously described a similar persistent association of α-syn with LAMP2A-positive lysosomes in PD models due to blockage in α-syn translocation inside lysosomes. These findings suggest, thus, a similar CMA blockage in the PD astrocytes at the receptor level. Also supportive of reduced α-syn degradation, western blot analysis confirmed a higher monomeric protein level of α-syn in PD astrocytes compared with controls (Figures 5D, 5E, and S7A). By using an antibody that detects specifically oligomeric α-syn, we were able to detect other pathogenic forms of α-syn in PD astrocytes, which were similar to those of PD postmortem brain tissue (Figure S7B).

17To investigate the contribution of the defect in CMA to the progressive accumulation of α-syn in PD astrocytes, we next performed a knockdown of LAMP2A using lentiviral-mediated short hairpin RNA (shRNA) targeting and silencing the LAMP2A spliced transcript (shLAMP2A), or an shRNA targeting the Luciferase gene (shLuc) as a control (Figure 5F). The shLuc control astrocytes displayed an expected low level of α-syn, whereas after shLAMP2A transduction, there was a statistically significant (p < 0.001) 2.5-fold increase in α-syn puncta, comparable to the levels observed in PD astrocytes (Figure 5G). Knockdown of LAMP2A did not change α-syn puncta levels in PD astrocytes, further suggesting defective CMA for α-syn in these cells. CMA activity was monitored using a photoactivatable CMA reporter, KFREQ-Dendra, in all astrocyte lines at 52 hr after photoactivation (Figures 5H and S6C). KFREQ-Dendra is present in the cytosol (diffuse fluorescent pattern) but as it is delivered to lysosomes via CMA it changes to a fluorescent punctate pattern. Ctrl astrocytes displayed these puncta, indicative of functional CMA, whereas the signal in PD astrocytes remained diffused in the cytosol, suggestive of an inactive CMA.

18Since PD astrocytes displayed higher levels of α-syn compared with Ctrl astrocytes, we next investigated possible differences in α-syn turnover in these cells. α-syn has previously been shown to undergo degradation both by the ubiquitin/proteasome system and by autophagy; therefore α-syn flux in the presence of lysosomal and proteasome inhibitors (leupeptin and lactacystin, respectively) was evaluated in Ctrl and PD astrocytes at 14 days (Figures 5I, 5J, and S7D). An increase of ∼40% in α-syn levels was found in Ctrl astrocytes after leupeptin treatment, while this increase was not found in PD astrocytes analyzed under the same conditions, indicating an impaired flux. No changes were found in either Ctrl or PD astrocytes after lactacystin treatment (Figures 5I, 5J, and S7D). These findings suggest major alterations in α-syn proteostasis due to poor degradation by lysosomal systems in PD astrocytes.

19Cells often respond to blockage in CMA by upregulating other autophagic pathways such as macroautophagy. However, altered macroautophagy has also been reported in the context of PD. To investigate the status of macroautophagy, the endo/lysosomal marker LAMP1, autophagosome marker LC3, astrocyte marker GFAP, and nuclear DAPI were used during ICC on all astrocyte lines at both 6 and 14 days. In Ctrl astrocytes, there was lysosomal LAMP1 staining in the perinuclear area and very few visible autophagic vacuoles (LC3-positive vesicles) at both 6 and 14 days (Figures 6A, 6B, and S6D). In PD astrocytes, as for LAMP2A, LAMP1-positive vesicles lost the preferable perinuclear distribution and were found throughout the entire cell (Figures 6A, 6B, and S6D). In addition, there was a marked increase in autophagic vacuoles starting as early as 6 days that continued increasing through the 14-day time point (Figures 6A, 6B, and S6D). Most of the accumulated LC3-positive vesicles in PD astrocytes did not co-localize with LAMP1 lysosomes (Figures S6E and S6F), suggesting that they were autophagosomes that persisted in these cells due to their poor clearance by lysosomes.

20In agreement with the immunofluorescence studies, western blot analyses detected higher basal levels of LC3-II in PD astrocytes compared with Ctrls (Figures 6C, 6D, and S7C). To monitor the autophagy flux and to gain insights into the mechanism behind the accumulated LC3-II levels in PD astrocytes, both Ctrl and PD astrocytes were treated with leupeptin and NH4Cl, inhibitors of lysosomal proteolysis, to inhibit LC3-II degradation. Under these conditions, PD astrocytes exhibited lower increase in LC3-II levels compared with controls, suggesting an impairment of the autophagy flux in PD astrocytes (Figures 6E and S7C). Moreover, we found higher p62 levels in PD astrocytes at baseline compared with controls, and lower flux ratio in the presence of inhibitor (Figures 6F–6H and S7E). Overall these findings suggest that reduced function in both autophagic pathways, CMA and macroautophagy, contribute to the altered α-syn proteostasis observed in PD astrocytes.

21Intracellular accumulation of α-syn has been shown to contribute to cellular toxicity in PD and to further disrupt functioning of cellular proteostasis systems (reviewed in ). We next investigated whether α-syn accumulation in PD astrocytes could be ameliorated by enhancing lysosomal activity. PD astrocytes were treated with a CMA activator (CA), which operates through the release of the endogenous inhibition of the retinoic receptor-α signaling pathway over CMA. Cells were treated with a concentration of 20 μM for 5 days and levels of α-syn were analyzed by immunofluorescence (Figure 7). LAMP2A-positive lysosomes in PD astrocytes treated with the CA (Figures 7C–7E) recovered the perinuclear distribution observed in Ctrl cells (Figure 7A) compared with when not treated (Figure 7B), suggesting reactivation of CMA in these cells. Consistent with higher CMA activity, CA-treated cells had significantly lower α-syn content than untreated cells (Figures 7C–7E).

22In addition, we treated PD astrocytes when in co-culture with Ctrl neurons (Figure 7F). Under untreated conditions, Ctrl neurons accumulate astrocytic α-syn and show signs of neurodegeneration. Interestingly, the treatment with CA cleared out α-syn not only in astrocytes, but also in vmDAns, partially restored neuron survival, and decreased the number of TH-positive cells with a degenerative morphology (Figures 7G–7L). These findings suggest that although multiple protein degradation pathways fail to efficiently degrade α-syn in PD cells, reactivation of one of these pathways, in our case CMA, is able to restore functional α-syn proteostasis.

23Astrocytes from three PD patients carrying the G2019S mutation in the LRRK2 gene and two healthy individuals were successfully generated using a previously published protocol and fully characterized. By implementing a patient iPSC-based co-culture model, in this study we describe a role for PD astrocytes in midbrain neuronal cell death. Specifically, in a 4-week co-culture system, we found morphological alterations resembling those of neurodegeneration, such as short and few neurites as well as beaded necklace-like neurites, typically observed in neurons upon transport alterations, and increased neuronal loss in Ctrl neurons co-cultured with PD astrocytes. We interpret these altered phenotypes as representing PD astrocyte-induced vmDAn neurodegeneration. An alternative explanation could be that PD astrocytes impinged on the differentiation and/or maturation of DAn progenitors in our iPSC-derived co-culture system. However, while we cannot formally rule out this possibility, two lines of evidence strongly argue against it playing a significant role in the phenotypes described here. First, we used vmDA neural differentiation cultures at 35 days of differentiation for our co-culture experiments. At this time, most vmDAns are already committed in fate (TH+/FOXA2+, see Figure 2A), but are still at a stage of maturation that does not compromise their survival upon cell dissociation and plating on top of the astrocyte cultures. Second, the numbers of vmDAns at different time points along the co-culture experiments showed progressive decline in co-cultures with PD astrocytes, but no significant changes when co-cultured with Ctrl astrocytes (Figure S2C). These results indicate that few new TH+ neurons are generated during co-culture, and further support that the decreased numbers of vmDAns observed upon co-culture with PD astrocytes are a consequence of vmDAn degeneration. Importantly, the altered phenotypes were specifically observed in the subpopulation of midbrain dopaminergic neurons, as numbers of MAP2+/TH− neurons did not change significantly upon co-culture with Ctrl or PD astrocytes. In accordance with this, it has been already shown that α-syn toxicity was responsible for nigrostriatal neuronal cell death in midbrain cultures, a relevant finding regarding the particular vulnerability of nigral neurons in PD. However, it remains to be tested whether, in prolonged culture, PD astrocytes also impair the survival of TH− populations.

24Postmortem brain tissue of PD patients revealed α-syn accumulation in astrocytes. It has been previously described that astrocytes accumulate neuronal-derived α-syn as a mechanism of neuroprotection. Indeed, in our study we found that Ctrl astrocytes accumulate α-syn when co-cultured with PD neurons and partially rescued the morphological phenotype of neurodegeneration and clearance of neuronal α-syn. This behavior suggests a neuroprotective effect via inflammatory-mediated activation of the Ctrl astrocytes. In addition, by using a CRISPR/Cas9 gene-edited cell line tagging the endogenous SNCA locus with an FLAG tag, our results reveal that PD astrocytes also accumulate and transfer α-syn to the surrounding neurons, suggesting that astrocytes actively contribute to the distribution of α-syn.

25Taking into account that our PD astrocytes come from patients carrying the LRRK2 G2019S mutation, we investigated whether disease-specific phenotypes related to the mutation were present. The α-syn accumulation in our co-culture system indicated a disruption in the way α-syn is usually degraded in PD astrocytes. Degradation of α-syn has been shown to occur by both proteasome and autophagic pathways, and conversely, high levels of α-syn have been demonstrated to be toxic for both systems. Here we found that lysosomal degradation of α-syn was severely inhibited in PD astrocytes. We have previously described in PD neurons that one of the early events in the dysfunction of the proteostasis systems in these cells is the disruption of CMA by mutant LRRK2 binding to the LAMP2A, thus causing the accumulation of α-syn. Here we demonstrate that CMA is also altered in PD astrocytes and confirm that α-syn degradation by CMA in these cells was almost completely abolished.

26The increase in intracellular levels of α-syn, due to its poor degradation in PD astrocytes by CMA, may contribute to precipitating malfunctioning of other proteostasis mechanisms, such as the proteasome and macroautophagy. In fact, we demonstrated that macroautophagy was also markedly impaired in these cells, by showing higher basal levels of autophagic vacuoles (LC3-II) and the autophagic cargo p62, and reduced autophagic flux (for both LC3-II and p62). The lower co-localization between the autophagosomal and lysosomal markers observed in PD astrocytes suggests that the reduced autophagic flux is due to a defect in autophagosome/lysosome fusion, similar to that previously described in PD neurons.

27Taking into account the coordinate functioning of the proteolytic systems, and the fact that CMA disruption seems to occur early during the development of PD pathology, we attempted to restore normal α-syn proteostasis by enhancing CMA activity. Our findings in cells treated with a chemical activator of CMA suggest that upregulation of CMA is still possible in these cells and that this intervention is sufficient to return levels of α-syn close to those in Ctrl cells. Although α-syn was cleared, restoration during a co-culture with Ctrl neurons was only partial in terms of neurite length and number, suggesting that the neurodegeneration observed could also be due to other non-α-syn-related factors secreted by PD astrocytes.

28Overall, our findings propose a specific role for astrocytes in mediating dopaminergic cell death during PD. PD-specific phenotypes specifically related to dysfunctions in the pathways of protein degradation have been observed in PD astrocytes and not in Ctrl astrocytes. Dysfunctional CMA, progressive α-syn accumulation, and glia-to-neuron transfer found in our PD astrocytes are all aspects that can compromise neuronal survival during PD pathogenesis. Future studies will identify whether additional factors other than α-syn are being secreted by (or lacking in) PD astrocytes, and thus contributing to triggering vmDA neuronal cell death. iPSC-based technology allows for the proper recapitulation of patient-specific disease-related phenotypes, which will aid in the discovery of new therapies.

29Experimental procedures are also provided in Supplemental Information.

30The parental iPSC lines used in our studies were previously generated and fully characterized. The generation and use of human iPSCs in this work were approved by the Spanish competent authorities (Commission on Guarantees concerning the Donation and Use of Human Tissues and Cells of the Carlos III National Institute of Health). iPSCs were differentiated into astrocytes following a previously published protocol. See Supplemental Information for more details.

31Four different iPSCs, two PD (SP12 and SP13) and two Ctrl (SP11 and SP11#4), were differentiated into dopaminergic neurons using a combination of two previously published protocols for midbrain induction. Detailed methods are provided in Supplemental Information.

32ICC on cell cultures was performed as described in Supplemental Information.

33Statistical analyses of the obtained data were performed using two-tailed unequal variance Student’s t tests and ANOVA (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001), and the mean and standard error of the mean were plotted using Prism (Mac OS X). Number of independent experiments (n) is indicated in each figure legend.

34Conceptualization, A.C. and A.R.; Methodology, I.F., J.P.M., A.Z., A.M.C., and J.S.; Investigation, Y.R.P, A.D., G.C., C.C., M.P.-E., M.G., I.F.-C., J.P., and A.F.; Validation, A.C. and A.R.; Writing – Original Draft, A.D.; Writing – Review & Editing, A.C. and A.R.; Visualization, A.D.; Resources, A.Z., A.M.C., J.S., E.T., A.C., and A.R.; Funding Acquisition, A.C.; Supervision, A.C.